Reading Results

Use result(options?) to select the output shape that best fits your application code.

Choose A Result Shape

import { ObjectCheck } from '@samatawy/checks';

const check = await ObjectCheck.for({
  profile: {},
  tags: []
}).check(input => [
  input.required('profile').object().notEmpty(),
  input.required('tags').array().notEmpty()
]);

console.log(check.result({ language: 'en' }));
console.log(check.result({ flattened: true, language: 'en' }));
console.log(check.result({ nested: true, language: 'en' }));
console.log(check.result({ validated: 'partial', language: 'en' }));
console.log(check.result({ validated: 'strict', language: 'en' }));
console.log(check.result({ raw: true, nested: true, flattened: true, language: 'en' }));
Copy

Use these options as a rule of thumb:

• language for the merged nested result tree
• catalog when you want result formatting to resolve coded messages through a specific ResultCatalog instead of ResultCatalog.global
• flattened: true when you only need message arrays such as errors
• nested: true when you want an input-shaped projection under input
• validated: 'partial' when you want a cloned value with invalid fields or items removed while valid siblings stay
• validated: 'strict' when any invalid descendant should remove the whole parent branch from the validated output
• raw: true when you also want the merged internal result tree exposed under raw

validated is built from the normalized input value after coercions such as trim() or tolerant boolean and number parsing. It is intended as a runtime-safe filtered clone, not as inferred compile-time typing.

Use Coded Results Only When Needed

If you need stable codes and translation catalogs, read coded-results.md.

That page covers:

• ResultCatalog.global
• separate ResultCatalog instances
• code on results
• fallback behavior when a translation is missing